Dear SPMers,
we're using a blocked design (blocks around 6 s, separated by resting
blocks of 4, 6, or 8 s, four experimental conditions plus target
stimuli). Within a block, 20 visual stimuli of the same condition are
presented rapidly (average SOA = 300 ms). Randomly, individual stimuli
are replaced by target stimuli (targets occur within 50% of blocks). I
have run four different 1st level model specifications in 11 of our
subjects and compared the individual results by eye.
The models were:
i) event-related analysis of all conditions
ii) blocked analysis of all conditions (except targets which were
modeled event-related)
iii) event related analysis of all conditions with having model
interactions (Volterra) as regressors in the design matrix, but SPM{T}
maps as output (only main effects, as in i) (I am not sure if this is
reasonable)
iv) event related analysis of all conditions with having model
interactions (Volterra) as regressors in the design matrix and
calculating an SPM{F} map for the first condition [1 0 0 0 0 0; 0 0 0 0
1], i.e. voxels active to cond1 and cond1*cond1. (I did it only for the
first condition cause I am not yet very familiar with all those
interaction regressors)
Results of i) and ii) look very, very similar and look somewhat better
than those of iii) and iv) (main effects at least). Comparing cond1 in
iii) and iv) does not indicate superiority for one approach. Personally,
I’d like to go for either i) or ii).
My questions are:
-is it – like in i) – okay to not model interactions in an event-related
analysis, even if there are 3 stimuli per second? To me, the SPMs of i)
look pretty reasonable. However, I read in the SPM list that I might run
into trouble if SOA is shorter than 1- 1.5 s and I do not account for this.
-Which way of model specification am I advised to use (from a
theoretical point of view)?
-In this context, I read in the SPM list about modeling the temporal
derivative of the HRF. Is that related to problems of interactions among
stimuli? How? Should I model derivatives?
I’d be grateful if you could help me with more insights.
Thank you very much,
Jens
--
Jens Kronschnabel, Ph.D. student
Department of Child and Adolescent Psychiatry, University of Zurich
Neumünsterallee 9, CH-8032 Zürich, Switzerland
+41 43 556 4081 [log in to unmask]
--
|